Identification of a regulatory domain controlling the Nppa-Nppb gene cluster during heart development and stress

Cell type-specific gene expression is regulated through regulatory DNA elements such as enhancers and repressors that function in strictly context-dependent manners. This context includes sequence- and epigenetic signature-dependent binding of complexes of transcription factors and co-factors at selective locations in the DNA and the three-dimensional (3D) conformation of the chromatin that brings these locations together. There is a paucity in the understanding of these mechanisms as only a few regulatory elements have been identified that drive gene expression during development and under pathological conditions in vivo (Kathiriya et al., 2015; de Laat and Duboule, 2013; Sanyal et al., 2012). To gain insight into these mechanisms, we studied the regulation of the Nppa-Nppb gene cluster during cardiac development and hypertrophy.

In the embryo, natriuretic peptide type A (also known as atrial natriuretic factor, ANF) and type B (also known as brain natriuretic peptide, BNP), encoded by Nppa and Nppb, respectively, are expressed in the atrial and ventricular myocardium. Both genes are also expressed in the adult heart (Cameron et al., 1996; Houweling et al., 2005) but Nppa is downregulated in the ventricles around birth to become restricted to the atria and the ventricular conduction system (Moorman and Christoffels, 2003). Both genes are strongly induced in the ventricles by stress and heart failure, and have become reliable clinical and experimental biomarkers for the severity of hypertrophy and heart failure (Guo et al., 2011; Sergeeva et al., 2014; Troughton et al., 2014). Moreover, Nppa is an important marker of early developmental malformations leading to congenital heart defects (Bruneau, 2011). Although the clinical relevance of Nppa and Nppb expression has led to studies on the regulation of Nppa and Nppb expression, the understanding of this complex process is still very limited.

Nppa and Nppb have descended from an ancestral CNP-3 gene by duplication and variation (Inoue et al., 2003). They are positioned at 10 to 15 kb distance from each other in the mammalian genome. Analysis of other gene clusters, such as Iroquois and Hox, revealed that multiple genes within a cluster have similar expression patterns and are regulated by shared regulatory sequences (Nolte et al., 2013; Tena et al., 2011). These features may underlie the evolutionary stability of such clusters. Therefore, we hypothesize that Nppa and Nppb are co-regulated by shared regulatory elements and transcriptional mechanisms.

Previously, it has been shown that the proximal Nppa promoter responded to stress in vitro but not in vivo (Knowlton et al., 1995), suggesting involvement of distal regulatory elements required for stress-induced Nppa expression. The region spanning from −141 kb to +58 kb relative to Nppa contains the transcriptional control information sufficient for both the developmental expression pattern and stress response of Nppa and Nppb (Sergeeva et al., 2014). Several regulatory elements located 20-37 kb upstream of Nppa were able to drive Nppa-like expression patterns during development (Matsuoka et al., 2014; Warren et al., 2011). One fragment induced reporter gene reactivation during cardiac hypertrophy. However, this fragment is located outside the −27/+58 kb region sufficient for the Nppa stress response (Horsthuis et al., 2008). The promoter regions of human (Majalahti et al., 2007), rat (He et al., 2001) and mouse (V.M.C., unpublished observations) Nppb in transgenic mice showed only very limited capacities to drive reporter expression in the normal and the stressed heart, and the developmental activity of these regions has not been described. Therefore, we have an incomplete understanding of the coordinated action of developmental and stress-induced regulatory elements.

Here, we characterized the spatial organization of the Nppa-Nppb cluster in the normal and the stressed heart using 4C-seq (van de Werken et al., 2012). The regulatory potential of the contacting sequences within the cluster were analyzed with chromatin immunoprecipitation (ChIP) for the enhancer-associated acetylated histone H3K27 (H3K27ac) and RNA polymerase II (Pol2). To identify regulatory elements involved in co-regulation of Nppa and Nppb, we made transgenic mice carrying modified bacterial artificial chromosomes (BACs) with reporter genes for both Nppa and Nppb, allowing the simultaneous monitoring of their dynamic expression patterns. Our study revealed that developmental and stress-induced expression of Nppa and Nppb is regulated by the interplay between shared developmental and non-shared stress-induced regulatory elements and dynamic epigenetic states within a 60-kb regulatory domain centered around Nppb.

To assess the topology of the Nppa-Nppb cluster and to define its potential regulatory domain, we characterized the interactions between different sequences (contact profiles) in the vicinity of Nppa and Nppb using high-resolution 4C-seq (Fig. 1A,B). Regulatory units are often flanked by CTCF-associated sites, where CCCTC-binding factor (CTCF) bound to DNA promotes DNA looping for proper regulation of gene expression (Lee and Iyer, 2012). The two closest CTCF-associated sites, located 41 kb upstream and 7 kb downstream of Nppa, were used as viewpoints. All regions contacting the viewpoints were sequenced and mapped to the genome. The median (±30% interval) of contact intensities (black curves and the gray bands, respectively, in Fig. 1B) show that the upstream CTCF-associated site (CTCF left) interacts with the downstream CTCF-associated site (CTCF right). Although vice versa interactions are less pronounced, these data indicate that the CTCF-associated sites encircle both Nppa and Nppb (Fig. 1A,B). The convergent orientation of these CTCF sites supports potential loop anchoring at the domain boundaries (Rao et al., 2014; de Wit et al., 2015). Moreover, both CTCF-associated sites interact with the Nppa promoter and another region within the locus (black arrowheads), showing possible involvement in the regulation of Nppa expression.

The data obtained from the promoter viewpoints indicate that interactions of both Nppa and Nppb are enriched between the CTCF-associated sites (Fig. 1B, second and third color-coded heatmaps from the top). Nppb lies in the center of these interactions with limited contacts beyond the CTCF sites. Although Nppa interactions are more widespread, most of them are concentrated within the 60-kb region flanked by the CTCF sites. Moreover, Nppa contacts are unidirectional and oriented towards Nppb. Interactions between the promoters of Nppa and Nppb are difficult to assess because the distance between these promoters is close to the spatial resolution of the 4C-seq analysis. Nppa and Nppb both interact with a region located 35 kb upstream of Nppa (black arrowheads). The interactions within the Nppa-Nppb locus are not heart specific, as they are also present in the liver (Fig. 1C; Fig. 2B), where Nppa and Nppb are not expressed. In both tissues, interactions of the CTCF-associated sites with the surrounding sequences are unipolar, contacting mostly the Nppa-Nppb locus (Fig. 1C, arrows). The frequency of these contacts is higher in the heart than in the liver. By contrast, interaction frequencies on the opposite side of the viewpoints are similar in both tissues (Fig. 1C, blue/red curves).

We next assessed whether the other genes in the vicinity in a 200-kb DNA region are co-expressed in the heart and respond to cardiac stress similarly to Nppa and Nppb, which would suggest co-regulation of these genes within the region. The expression levels of ten genes surrounding Nppa and Nppb were measured in adult ventricles by qPCR (Fig. 1D). Ventricular expression of Nppa decreases after birth (Horsthuis et al., 2008) to a level lower than that of Nppb. However, mRNA levels of Nppa were in large excess (7-170×) over that of the other genes in the locus. In the pressure-overloaded heart, 4 weeks after transverse aortic constriction (TAC) surgery, the heart weight (HW) to tibia length (TL) ratio and the levels of Nppa and Nppb were increased compared with control mice (Fig. 1E,F), whereas the expression of the other genes was not significantly changed (Fig. 1E). Therefore, among ten genes in this 200-kb DNA region, only Nppa and Nppb are highly expressed in the myocardium and induced after stress, suggesting that Nppa and Nppb do not share regulatory sequences with the surrounding genes. This analysis indicates that the sequences that regulate Nppa and Nppb lie within the 60-kb region flanked by CTCF-associated sites and form an Nppa-Nppb regulatory interaction domain. The conformation of this domain is not heart specific, although higher frequencies of interactions within the domain in the heart might contribute to the regulation of heart-specific Nppa and Nppb expression.

To locate discrete potential regulatory sequences in the 60-kb Nppa-Nppb regulatory domain in the fetal heart and during disease, we analyzed contacts of the Nppa and Nppb promoters with other sequences inside their interaction domain. We aligned cardiac ChIP-seq data for the CTCF sites with the 4C contact profiles generated from different heart regions and liver (Fig. 2A,B; Fig. S1). The bait regions in the promoters of Nppa and Nppb interact with each other as well as with the regions located between −35 kb and +7 kb relative to Nppa (red regions in the heatmaps). Notably, the patterns of these interactions are similar in the atria, fetal and adult ventricles, stressed ventricles, and even in the liver. Paired analysis of the adult and stressed ventricles, as well as the ventricles and the liver revealed that the interaction frequencies between the Nppb promoter and contact sites are similar (Fig. 2C). Taken together, the conformation of the Nppa-Nppb locus is permissive, i.e. not different between stages, tissues or conditions.

We assessed the distribution and levels of H3K27ac, a chromatin modification associated with active enhancers, and the occupancy of Pol2, a mark of active promoters and enhancers. H3K27ac modification and Pol2 association are heart specific within the Nppa-Nppb interaction domain (gray area and magenta arrowheads) in a 200-kb DNA region (Fig. 3A). To assess whether the levels of H3K27ac and Pol2 occupancy are altered in the stressed heart, we performed ChIP-qPCR analysis in the ventricles of control and TAC mice 4 weeks after the operation. H3K27ac and Pol2 marks were enriched around Nppa and Nppb as well as in the region −40/−30 kb upstream of Nppa in control hearts (Fig. 3B,D). These data correspond to the ENCODE data for H3K27ac and Pol2 in the heart (Stamatoyannopoulos et al., 2012) (Fig. 3C). H3K27ac association was significantly different in the TAC hearts at many sites, whereas Pol2 occupancy changed much less. Increased H3K27ac was observed in the Nppa and Nppb promoters as well as at −18 kb, +2.3 kb and +15 kb relative to Nppa. Interestingly, the acetylation levels at the region 35-40 kb upstream of Nppa were decreased in TAC hearts compared with controls. Decreased Pol2 association at −37 kb relative to Nppa accompanied these changes.

Taken together, these data indicate global changes in H3K27ac association in the Nppa-Nppb regulatory domain after stress, with increased levels close to the genes, and decreased levels of both H3K27ac and Pol2 association in the conserved upstream region previously associated with developmental regulation of Nppa expression (Horsthuis et al., 2008).

To test potential enhancer activity of parts of the region contacting the Nppa and Nppb promoters, individual fragments covering the region from −27 kb to +58 kb relative to Nppa were tested in transient transfection assays. Two fragments at −22 kb and at −11 kb relative to Nppa showed enhancer activity (not shown) and a moderate increase of H3K27ac levels at these regions after TAC (Fig. 3B). We generated transgenic mice with these fragments coupled to the Nppa basic promoter and EYFP reporter. Both −22-prNppa-EYFP and −11-prNppa-EYFP reporter constructs provided the expression pattern in the fetal heart typical of Nppa (Fig. 4A,D). Deletion of the −22 kb fragment from BAC336-KL did not impair the fetal expression pattern of Luciferase (reporting for Nppa activity) (Fig. 4G). These data suggest the presence of several redundant developmental Nppa enhancers.

To test the requirement of potential regulatory fragments in the context of the Nppa-Nppb locus, we generated a series of overlapping and truncated BAC double-reporter transgenic mice and compared the developmental profile and stress response of the reporter genes with those of the original BAC336 (Figs 5 and 6). The proximal promoter of Nppa present in all BACs contains sufficient information to drive robust atrial expression in embryonic and adult hearts (Horsthuis et al., 2008). Therefore, we report the expression levels of both Luciferase/EGFP (Nppa reporter) and Katushka (Nppb reporter) in fetal and adult ventricles relative to the levels of atrial Luciferase/EGFP in each BAC mouse line. Two independent transgenic lines of BAC336-KL(I) and of BAC336-KL-Δdown(II) showed similar dynamic ventricular expression of Luciferase and Katushka (Fig. S2). Therefore, we further compared the expression of the reporters between different BAC transgenic lines.

The expression patterns of Katushka and Luciferase in BAC336-KL(I) recapitulated the expression pattern of Nppb and Nppa, respectively (Fig. 5A,B). Deletion of the downstream sequences in BAC336-KL-Δdown(II) or intergenic sequences in BAC336-KL-Δinter(III) did not alter the expression patterns of Luciferase and Katushka (Fig. 5B; data not shown). By contrast, deletion of the upstream sequences in BAC337-KE(IV) resulted in undetectable Katushka and reduced EGFP expression (Fig. 5B, lower panel). Quantification of the fetal ventricular activity of Luciferase/EGFP and Katushka by qPCR in the transgenic BAC lines confirmed these results (Fig. 5C,D). These findings suggest the presence of fetal ventricular enhancers of Nppa and Nppb in the distal upstream region (>27 kb upstream of Nppa).

Deletion of the downstream region in BAC336-KL-Δdown(II) and the intergenic region in BAC336-KL-Δinter(III) did not impair Nppa-characteristic postnatal downregulation of Luciferase (Fig. 5C). Similarly, the moderate decrease of Nppb after birth was recapitulated by the BAC transgenics lacking the downstream or intergenic region, although the postnatal decrease in reporter expression was stronger (Fig. 5D). Adult ventricular expression of Katushka was reduced in BAC337-KE(IV) and BAC336-KL-Δup(V) in which sequences more than 27 kb and 17 kb upstream of Nppa, respectively, were deleted. These data indicate that adult ventricular enhancers of Nppb are located upstream of the gene.

We next determined whether Luciferase/EGFP and Katushka are controlled by the same regulatory elements (Fig. 5E). Adult ventricular expression of Luciferase/EGFP and Katushka were significantly reduced in all the lines with the modified BACs (II-V) compared with those of Luciferase and Katushka in BAC336-KL(I), respectively. This analysis suggests either the presence of multiple regulatory regions upstream, in between and downstream of both genes, or an influence of the disturbed conformation of the locus resulting from the deletions. However, because of the multiple copies of the BACs integrated in the genome, interpretation of the 4C-seq data from these BAC lines was not possible (data not shown). The levels of Katushka in BAC336-KL-Δdown(II), BAC336-KL-Δinter(III) and BAC336-KL-Δup(V) were consistently higher than those of Luciferase, mimicking the ratio of endogenous Nppa/Nppb (Fig. 5E). However, in BAC337-KE(IV), which lacks the upstream region, the ratio between EGFP and Katushka expression levels in the ventricles was reversed and significantly different from that in BAC336-KL(I) (Fig. 5E, insert). These data indicate that the distal upstream region is required for adult basal ventricular activity of Nppb. In BAC336-KL-Δup(V), ventricular Luciferase levels were again lower than Katushka levels, suggesting that adult ventricular Nppa expression is regulated by sequences located from −27 to −17 kb relative to Nppa.

Taken together, we conclude that prenatal ventricular expression of Nppa and Nppb are co-regulated by, possibly shared, distal upstream elements, but that postnatal expression involves different elements. Our data also suggest that the topology of the locus, influenced by the deletions, affects the regulation of the Nppa-Nppb cluster.

To test whether the two developmental enhancers located −22 and −11 kb upstream of Nppa contain stress-response elements, the −22-prNppa-EYFP and −11-prNppa-EYFP reporter mice were subjected to TAC surgery. EYFP expression levels were not induced in the ventricles 4 weeks after TAC surgery (Fig. 4B,C,E,F). Moreover, the levels of Nppa correlated with that of Luciferase in the ventricles of BAC336-KL-Δ(−22) mice after TAC (Fig. 4H,I). Therefore, the −22 kb and −11 kb regions, which contribute to the developmental expression of Nppa, are not involved in its stress-induced upregulation. Similarly, the levels of Katushka mRNA increased after TAC and mimicked that of Nppb, showing that the −22 kb fragment is not involved in stress-induced upregulation of Nppb (Fig. 1H).

To identify stress-response region(s) of Nppb and Nppa, we analyzed the induction of reporter gene expression in the other BAC transgenic mice after TAC. The induction of Katushka and Luciferase mRNA in BAC336-KL(I) mice after TAC recapitulates the induction of Nppb and Nppa, respectively (Sergeeva et al., 2014). We compared the response of the reporters in the reference BAC336-KL(I) line with that in other transgenic lines that lack regions contacting the Nppa/Nppb promoters (Fig. 6, Figs S3, S4 and Table S1). This comparison allowed us to assess the necessity of critical regulatory regions for Nppa and/or Nppb induction.

Deletion of the region downstream of Nppa in BAC336-KL-Δdown(II) and of the 12-kb intergenic region in BAC336-KL-Δinter(III) did not reduce the TAC-induced response of Katushka and Luciferase in the ventricles compared with BAC336-KL(I) (Fig. 6). Although the response of Luciferase in one of two independent lines of BAC336-KL-Δdown(II) is significantly stronger than that in BAC336-KL(I), we interpret both lines as containing sufficient regulatory information for reporter induction (Fig. S5). These data reveal that the potential enhancers are located in the regions upstream of Nppb, in agreement with the loss of the stress response of Luciferase in RP23-139J21(VI) (Fig. 6, Fig. S3). Deletion of the sequences upstream of −27 kb in BAC337-KE(IV) and of −17.2 kb in BAC336-KL-Δup(V) relative to Nppa, resulted in reduced reporter gene activity in the adult ventricles (Fig. 5E). Nevertheless, EGFP expression (Nppa reporter) in BAC337-KE(IV) was upregulated after TAC, whereas analogous expression of Luciferase in BAC336-KL-Δup(V) was not (Fig. 6, Fig. S3). By contrast, the response of Katushka (Nppb reporter) to stress was decreased in both lines compared with that in BAC336-KL(I), although these differences did not reach statistical significance (Fig. 6; Figs S3 and S4; lines IV and V). Therefore, sequences located >27 kb upstream of Nppa might contribute to stress-induced Nppb expression. However, sequences crucial for Nppa upregulation are located in the region between −27 and −22 kb upstream of Nppa, implying that different mechanisms regulate Nppa and Nppb.

Because the promoters of Nppa and Nppb were shown to interact with each other (Fig. 1B), we deleted the 0.6-kb Nppb promoter in BAC336-KL(I) to assess its contribution to stress-induced Nppa expression. The response of Luciferase to stress in BAC336-KL-ΔprNppb(VII) mice was completely abrogated compared with that in BAC336-KL(I) (Fig. 6; Fig. S3). These data suggest that the Nppb proximal promoter is required for Nppa upregulation during stress. A 2-kb Nppb promoter fragment exhibits enhancer activity when tested in neonatal rat ventricular myocytes (NRVM) stimulated with phenylephrine (Fig. S6). However, this 2-kb Nppb promoter fragment coupled to the 0.7-kb Nppa promoter-Luciferase was not sufficient to induce Luciferase expression in the stressed heart (Fig. 6; Fig. S3). These data indicate that the Nppb promoter contains regulatory information contributing to, but not sufficient for stress-induced expression of Nppa in vivo.

Taken together, we conclude that Nppa stress-response enhancers are different from its developmental regulatory sequences. Induction of Nppa after stress is dependent on the Nppb promoter and the sequences located in the region from −27 to −22 kb relative to Nppa. Nppb basal ventricular activity, as well as stress-induced expression, is regulated by the sequences located >27 kb upstream of Nppa. Therefore, stress-induced reactivation of Nppa and Nppb is regulated by different mechanisms.

The 4C-seq analysis of the chromosome conformation of the Nppa-Nppb locus revealed interactions occurring between several potential regulatory regions (Fig. 2) within a 60-kb domain flanked by CTCF-associated sites. Inside this domain, −41/−30 kb, −15/−11 kb (containing Nppb) and −0.7/+4 kb (containing Nppa) regions relative to Nppa contain regulatory elements as predicted by the EMERGE enhancer prediction tool (van Duijvenboden et al., 2016) (Fig. 7A). This prediction correlates well with the developmental and stress-induced enhancer regions identified in our functional studies using transgenic mice (Fig. 7A). Taking the dynamic ventricular expression of Nppa and Nppb into account (Fig. 7B), we constructed a functional 3D model of the Nppa-Nppb locus (Fig. 7C). In this model, a chromatin loop anchored by the convergent CTCF sites encloses Nppa, Nppb and their regulatory regions. Although the promoters of Nppa and Nppb interact with each other and with the sequences located up to 40 kb upstream of Nppa, different sets of regions are involved in the dynamic regulation of expression of these two genes. The region >27 kb upstream of Nppa, presumably the −41/−30 kb element, regulates embryonic, adult and stress-induced expression of Nppb. Supposedly, dynamic levels of H3K27ac association with this region as well as with the promoter of Nppb might contribute to the relatively modest changes of Nppb expression in the postnatal and stressed heart. By contrast, Nppa ventricular expression dramatically decreases after birth and increases after stress (Fig. 7B). Therefore, the Nppa promoter is modulated by several regulatory elements, some of which are shared with Nppb. Postnatal downregulation of Nppa might be governed by the sequences located −27 to −22 kb relative to its start, but during stress Nppa additionally functionally interacts with the Nppb promoter, which is required, but not sufficient, for Nppa upregulation. The proposed model indicates that the regulatory information of the gene cluster is distributed across the locus rather than present specifically in one or a few shared discrete enhancers, and predicts that the 3D conformation and changes in epigenetic state of the locus contribute to the regulation of Nppa and Nppb during development and disease.

The expression levels of Nppa and Nppb are almost exclusively heart specific (Gardner et al., 1987) and were by far the highest of the ten genes in the 200-kb DNA region around the cluster. The selective cardiac-specific expression and coordinated induction of Nppa and Nppb during hypertrophy suggest the presence of regulatory elements in the locus selectively targeting these genes.

Promoters and their regulatory elements interact with each other within topologically associated domains (TADs). TADs are separated by boundary regions enriched for convergent CTCF-associated sites (Ong and Corces, 2014; Rao et al., 2014; de Wit et al., 2015). Our 4C-seq analysis indicates that the CTCF-associated sites flanking Nppa/Nppb contact each other, encircling the gene promoters and most of their interaction partners. Although we have not shown the requirement of the CTCF sites for Nppa/Nppb regulation, we suggest that CTCF protein complexes anchor the loop using the convergent CTCF-binding sequences and maintain the stability of the regulatory domain. Moreover, the upstream CTCF site might also be associated with enhancer elements, because this region contacts the Nppa promoter and is enriched for H3K27ac, H3K4me1, p300 and DNaseI hypersensitive sites (HSs), indicative of potential enhancer activity. Furthermore, the topology of the locus was similar in atria, ventricles and liver, and in fetal and adult stages, qualifying the conformation of the Nppa-Nppb locus as permissive, where the contacts are pre-formed and not cell type dependent (de Laat and Duboule, 2013).

TAD boundaries are static across species and cell types (Dekker et al., 2013). When comparing the organization of the Nppa/Nppb-containing region in the mouse heart with that in other tissues we found that the 60-kb regulatory domain is a structural subunit of a larger ∼600-kb TAD. The upstream border of this TAD in mouse cerebellar cortex and embryonic stem cells coincides with the boundary we observed with 4C-seq and is relatively sharp (Dixon et al., 2012). Downstream of Nppa, interactions gradually decrease by the position of Clcn6 (20 kb downstream Nppa), as observed by 4C-seq, suggesting a much less strict boundary between the 60-kb domain and the remainder of the TAD. This (sub)domain organization suggests the establishment of a rigid chromatin organization already at very early stages in mouse development. Hi-C data from several human cell lines, performed with higher resolution compared with the mouse 4C-seq data, indicates the presence of a 100-kb domain similar to the mouse 60-kb domain, i.e. including only NPPA, NPPB and a conserved region upstream of both genes (Rao et al., 2014).

Previously, genome-wide association studies have identified genetic variants influencing the concentration of natriuretic peptides in the blood of heart failure patients. Two single nucleotide polymorphisms, rs5065 in an exon of NPPA (Vassalle et al., 2007) and rs1023252 in an intron of CLCN6 (Del Greco et al., 2011), showed association with NT-proBNP (N-terminal of the prohormone brain natriuretic peptide) levels in patients with cardiac dysfunction, indicating that these variants can affect the function of regulatory sequences. Genetic variants associated with hypertension (Flister et al., 2013) are positioned across a larger 300-kb region (from upstream of NPPB to downstream of AGTRAP). Because blood pressure is influenced by multiple genes within this region (MTHFR, CLCN6, NPPA, NPPB), these data suggest that the variants influence the expression of several genes within the TAD, including NPPA and NPPB. As these variants are in linkage disequilibrium, it is not yet known which variants influence the expression of NPPA and NPPB. Our data suggest that variants that influence NPPA/NPPB expression, and thus predispose to hypertension, are localized within the 60-kb regulatory domain.

Three natriuretic peptide genes developed during vertebrate evolution by duplication of a common ancestral gene. Mammals retain Nppa and Nppb, conserved among species and positioned at a distance of only 15 kb from each other in the genome (Inoue et al., 2003). Several studies of clustered genes describe coordinated regulation by shared regulatory sequences (Nolte et al., 2013; Tena et al., 2011). Given the highly similar spatiotemporal expression patterns of Nppa and Nppb, we hypothesized that these clustered genes might also share regulatory elements.

Previously, the region from −27 to +58 kb relative to Nppa was found to contain regulatory elements required for the stress-response of Nppa, whereas upstream sequences are responsible for its developmental activity (Horsthuis et al., 2008; Sergeeva and Christoffels, 2013). Three potential enhancer elements at −35, −31 and −21 kb relative to Nppa were identified based on their ability to bind Nkx2-5 in vivo. These elements, when combined, were able to drive correct prenatal ventricular expression, but they could not activate the lacZ reporter in the ventricles after TAC (Warren et al., 2011). A 650-bp conserved element at −37 kb upstream of Nppa was able to induce Luciferase expression in the ventricles of transgenic mice after TAC (Matsuoka et al., 2014). However, this element is located outside the region required for stress responsiveness of Nppa. These data suggest the presence of multiple enhancers regulating Nppa, and possibly Nppb, during disease. Our analysis of BAC transgenic mice revealed that Nppa and Nppb share some developmental enhancers located >27 kb upstream of Nppa. The highly conserved 10-kb DNA region located at −40/−30 kb of Nppa contains many DNaseI HS sites, sites occupied by cardiac transcription factors, p300, and histone modifications associated with gene regulation. This region contacts the promoters of both Nppa and Nppb and might represent a fetal ventricular enhancer region of these genes. Furthermore, this region might be involved in the basal and stress-induced expression of Nppb in the adult heart. Therefore, we suggest that the 650-bp fragment (Matsuoka et al., 2014) located within the −40/−30 kb region is one of the distal stress-response enhancers of Nppb, not Nppa. In support of this notion, we observed very strong EGFP (reporting for Nppa) upregulation in the stressed ventricles of BAC337-KE(IV) mice lacking the entire −40/−30 kb region.

We observed that the mouse Nppb promoter was required for basal Nppb expression in the fetal and adult heart, but not required for its stress induction. Furthermore, the Nppb promoter induced Luciferase expression in neonatal rat ventricular myocytes stimulated with phenylephrine, and was necessary, but not sufficient, for Nppa stress induction in vivo. Two previous studies found that the 408-bp human and 534-bp rat Nppb promoters were able to stimulate expression to some extent in myocardial infarction (MI) or angiotensin II-induced hypertrophy, respectively (He et al., 2001; Majalahti et al., 2007). Together, these data suggest that the promoter of Nppb might be part of a composite stress-inducing enhancer for Nppa, acting as a scaffold attracting other regulatory DNA elements and transcription factors for combinatorial enhancer activity in the context of the proper chromatin conformation and histone modifications. Because the randomly integrated BACs are not suitable for in vivo 4C-seq analysis, this hypothesis has to be tested by deletion of the region from the mouse genome using genome-editing technologies.

Although BACs usually provide an accurate expression pattern of the integrated transgene, they have some weaknesses related to the random site of integration, copy number and integrity. Therefore, we confirmed proper reporter gene expression in two independent founder lines for some of the BACs, and analyzed several overlapping BACs to confirm the results of individual transgenic lines. Taking the activities of the reporter genes in all mouse lines into account, we conclude that potential Nppa stress-response elements are located in the region from −27 to −22 kb relative to Nppa. Although ChIP-seq data sets indicate that this region does not bind cardiac transcription factors, part of this region interacts with the promoter of Nppa. Together, these data show that the clustered Nppa and Nppb genes share their developmental enhancers, but not their stress-response elements. Because fetal expression and stress-induced Nppa induction appear to be dependent on different enhancers, fetal expression and reactivation of the ‘fetal gene program’ during stress require different transcriptional mechanisms.

Histone modifications, such as acetylation or methylation of lysine residues of histone H3, are involved in transcriptional regulation. Such modifications have been associated with the development of cardiomyopathies in the adult heart (Mahmoud and Poizat, 2013). Demethylation of H3K9 at the promoter regions of NPPA and NPPB was associated with activation of their expression in failing human left ventricular myocardium, whereas H3K9ac was not changed and H3K27ac was modestly increased (Hohl et al., 2013), in accordance with our observations (Fig. 3B, red and green bars). Both Nppa and Nppb have been reported to recruit Pol2 de novo 4 days after TAC (Sayed et al., 2013). We did not observe significant changes in Pol2 binding at the promoter regions of Nppa and Nppb 4 weeks after TAC, presumably because phosphorylation of Pol2 at the transcription start site of Nppa is required for activation of its expression (Spiltoir et al., 2013). Although it is likely that the promoter regions participate in the induction of both genes, the mouse Nppa (Horsthuis et al., 2008; Knowlton et al., 1995) and Nppb (this study) promoters were not sufficient to provide the stress response. Therefore, distal elements must be required for the induction of Nppa and Nppb.

Genome-wide analysis of H3K27ac revealed that many loci are differentially associated with acetylated H3K27 during hypertrophy in mice (Papait et al., 2013). Close analysis of these data showed that H3K27ac levels were slightly increased at the promoter regions of Nppa and Nppb, but were not markedly changed in the −22 kb and −11 kb putative regulatory regions, in line with our findings. However, the −40/−35 kb region relative to Nppa, described above as a potential enhancer, was highly acetylated already in the normal heart (our data; Papait et al., 2013; Stamatoyannopoulos et al., 2012), and is likely to be involved in the adult expression of Nppb, which, in contrast to Nppa, is still expressed in the ventricles after birth. After TAC, the levels of H3K27ac in this region were significantly lower than in the normal heart. These variations in acetylation at specific loci might be involved in modulation of transcription factor binding (Reynolds et al., 2013) at enhancers for fine-tuning of Nppa/Nppb expression. Taken together, the data point to global epigenetic changes in the locus, which, in conjunction with multiple enhancers and in the context of a three-dimensional conformation convey the stress response to the transcriptional activity of Nppa and Nppb.

Animal care and experiments conform to the Directive 2010/63/EU of the European Parliament. All animal work was approved by the Animal Experimental Committee of the Academic Medical Center, Amsterdam, and was carried out in compliance with the Dutch government guidelines. Transgenic mice (Mus musculus) were generated and bred on the FVB/N background. See supplementary Materials and Methods for further details.

Preparation of 4C templates was described previously (Simonis et al., 2009). In short, adult mouse hearts and liver were isolated, dissociated and homogenized to obtain a single cell suspension. Chromatin was cross-linked with 2% formaldehyde, nuclei were isolated, and cross-linked DNA was digested with DpnII. Digestion was followed by proximity ligation, removal of cross-links, a secondary restriction digestion with Csp6I and a second proximity ligation. For all experiments, 200 ng of the resulting 4C template was used for the subsequent PCR reaction. After PCR, 4C templates were purified, bar-coded, mixed and sequenced simultaneously on Illumina HiSeq 2000 platform. Mapping and filtering of the sequence reads was carried out as previously described (van de Werken et al., 2012). See supplementary Materials and Methods for further details, including primer design, data analysis and statistics.

Left ventricles of the control and TAC hearts were dissected, ground in liquid nitrogen and cross-linked in 1% formaldehyde. Cross-linking was quenched, tissues were further dissociated and lysed. Cross-linked nuclei were sonicated under conditions established to yield an average fragment size of ∼300 bp. Antibodies were anti-H3K27ac (2 μg; Abcam, ab4729) and anti-Pol II (2 µg; Santa Cruz Biotech, sc-899X). Immunoprecipitation, washing, elution, and reverse cross-linking were performed as previously described (van den Boogaard et al., 2013). ChIP-qPCR was performed on a Roche LightCycler 480 System using Sybr Green monitoring. The ratio of ChIPed DNA normalized for input DNA was compared with that of a negative control region within Hprt and is presented as fold enrichment. See supplementary Materials and Methods for further details, including primer sequences.

Generation of BAC336-Katushka-Luciferase [BAC336-KL(I)] and BAC337-EGFP has been described previously (Horsthuis et al., 2008; Sergeeva et al., 2014). See supplementary Materials and Methods for further details of all constructs.

Wild-type or transgenic FVB/N male mice (8-12 weeks old) were subjected to transverse aortic constriction (TAC) as described (van Deel et al., 2011). All mice were weighed, sedated with 4% isoflurane and anesthetized with O₂/N₂ (v/v=1/2) containing 2.5% isoflurane. Buprenorphine (50 μg/kg) was injected subcutaneously for postsurgical analgesia. Age-matched unbanded littermates were used as controls. After 4 weeks, the ventricular apex was used for qPCR and the left ventricle was fixed for 4C and ChIP experiments. Ventricular tissue was dissected and apportioned for qPCR and in situ hybridization. See supplementary Materials and Methods for further details.

Infarction was created in 8- to 12-week-old BAC336KL and prNppb-prNppa-Luc male mice by permanent ligation of the left anterior descending coronary artery according to the protocol described previously (De Celle et al., 2004) with slight modifications. Mice were sedated with 4% isoflurane, analgesized subcutaneously with buprenorphine (0.068 mg/kg) and intubated. Anesthesia was maintained with 2% isoflurane in O₂ (1 l/min flow rate). After 1 week, the border zone of the operated mice and a piece of left ventricle of the control mice were dissected and used for qPCR analysis. See supplementary Materials and Methods for further details.

Dissected atrial and ventricular tissue of embryonic, control adult and TAC/MI-operated mice was dissected and snap frozen in liquid nitrogen and stored at −80°C. Total RNA was isolated using Trizol Reagent according to the manufacturer's protocol (Invitrogen). Three-hundred nanograms (fetal tissue) or 500 ng (adult tissue) of total RNA was used for reverse transcription with the Superscript II system (Invitrogen) and Oligo-dT as primers. Expression of different genes was assayed by quantitative real-time PCR using the LightCycler Real-Time PCR system (Roche Diagnostics). Values were normalized to Gapdh and Hprt expression levels. See supplementary Materials and Methods for further details, including primer sequences.

Non-radioactive in situ hybridization on sections was performed as described previously (Moorman et al., 2001). Embryos and hearts were fixed in 4% formaldehyde, embedded in paraplast and sectioned at 10 µm. The cDNA probes used were Nppa, Nppb, Luciferase, Katushka and EYFP. Images were acquired with the Zeiss Axiophot microscope. See supplementary Materials and Methods for further details.

Three fragments were PCR amplified from the mouse BAC336 clone containing the Nppa and Nppb loci (129 SvJ BAC library, Incyte, St Louis, MO, USA). The PCR fragments were subcloned into the lentiviral vector encoding the firefly luciferase reporter (pGreenFire Transcriptional Reporter Lentivector; System Biosciences, Mountain View, CA, USA). The supernatant from 293T cells containing the lentiviral particles was collected 48 and 72 h after transfection and concentrated by centrifugation in Amicon Ultra-15 Centrifugal Filter Unit (Millipore, UFC910008).

Rat neonatal ventricular myocytes (NRVMs) and fibroblasts were isolated from 1- to 2-day-old Lewis rats, infected with the lentiviral vectors and incubated for 16 h. Subsequently, the cells were exposed to 100 μM phenylephrine (Sigma, P6126-10G) for 48 h. Cell extracts and luciferase assay were performed according to the manufacturer's protocol using the Renilla Luciferase Assay System (Promega) and the Glomax-Multi detection system. See supplementary Materials and Methods for further details.

Fetal and adult ventricular Luciferase/EGFP and Katushka expression levels were normalized to the levels of fetal atrial Luciferase/EGFP (Fig. 2C,D). All data in bar charts are represented as mean value±s.e.m. The two-tailed Student's t-test was used to determine statistical significance (P<0.05) of the differences between the fetal [embryonic day (E) 17.5] and adult ventricular activity of the reporter genes, and the differences in normalized ventricular activity between different BAC transgenic lines and those of BAC336KL(I). In TAC and MI experiments, factor correction (Ruijter et al., 2006) was used to remove variability between experiments performed in several sessions. The two-tailed Student's t-test was used to determine statistical significance (P<0.05) of gene upregulation. Correlation coefficients were used to determine the relationship between the levels of Nppb and Katushka mRNA, and the levels of Nppa and Luciferase/EGFP mRNA, which represent the response of the reporter genes to stress. A t-test between the slopes of the regression lines was used to estimate the difference between the response of the reporter genes in two transgenic lines (Armitage et al., 2008). See supplementary Materials and Methods for further details.

We acknowledge the assistance and support of the Animal Research Institute of the Academic Medical Center (AMC), University of Amsterdam, in generating the BAC transgenic mouse lines. We thank Vincent Wakker for help in generating transgenic constructs and Corrie de Gier-de Vries for in situ hybridization.